Unless you are pre-transforming all of your vertex data, vertices should come into the vertex shader in object/model space. This is typically where the vertices are converted into screen space by transforming the vertex by the world, view, and projection matrices. If you want to perform operations on the vertices in world space, then either pass through the objects world transformation matrix in addition to a pre-multiplied world-view-projection matrix. Or pass in the world, view, and projection matrices separate from each other. You would then transform the vertex by the world matrix, perform all necessary operations on it, and then transform it by the view, and projection matrices before passing it through to the pixel shader.